A system monitors an individual for conditions indicating a possibility of occurrence of irregular heart events. A database includes a plurality of combinations of at least a first signature and a second signature. A first portion of the plurality of combinations is associated with a normal heartbeat and a second portion of the plurality of combinations is associated with an irregular heart event. A wearable heart monitor that is worn on a body of the patient includes a heart sensor for generating a heart signal responsive to monitoring a beating of a heart of the individual. The monitor further includes a processor for receiving the heart signal from the heart sensor. The processor is configured to analyze the heart signal using a plurality of different processes. Each of the plurality of different processes generates at least one of the first signature and the second signature. The plurality of different processes provide a unique combination including at least the first signature and the second signature for the generated heart signal. The processor compares the unique combination with the plurality of combinations in the database, locates a combination of the plurality of combinations that substantially matches the unique combination and generates a first indication if the unique combination substantially matches one of the first portion of the plurality of combinations and a second indication if the unique combination substantially matches one of the second portion of the plurality of combinations.
Legal claims defining the scope of protection, as filed with the USPTO.
2. The system of claim 1 further including an alarm for generating an alarm indication responsive to generation of the second indication.
3. The system of claim 1, wherein the database is located within the wearable heart monitor.
4. The system of claim 1, further comprising additional signatures comprising features of the heart signal, the features of the heart signal comprising at least one feature from the group consisting of, non-linear features and dynamical invariants.
5. The system of claim 1, further comprising additional signatures comprising features of the heart signal, the features of the heart signal comprising at least one feature from the group consisting of, chaotic features, catastrophe features, time domain features, frequency domain features, time-frequency domain features, and non-linear features.
6. The system of claim 5, wherein the topological features and the chaotic features comprise at least one of embedded dimensions of an attractor, fractional attractor dimensions, Correlation dimensions, Lyapunov exponent, Kolmogorov Entropy, Mutual Information, a proper delay for an attractor dimension, spectral density as a function of frequency, nonlinear chaotic measures, and a horizon for predictability.
7. The system of claim 5, wherein the time domain features comprise at least one of direct measurement of RR intervals, measurements of differences between RR intervals, a mean of all RR intervals, a standard deviation of all RR intervals, a square root of a mean of squares of differences between adjacent RR intervals and standard deviation of differences between adjacent RR intervals.
8. The system of claim 5, wherein the frequency domain features comprise low frequency band power spectral density features, high frequency band power spectral density features, very low frequency band power spectral density features, and ratio of low frequency band and high frequency band power spectral density.
9. The system of claim 5, wherein the time-frequency domain features comprise maximum amount of energy in a time window, minimum amount of energy in the time window, difference between maximum and minimum amount of energy between time windows, standard deviation between energy of time windows, total energy of a signal in a low frequency band, total energy of a signal in a high frequency band, average energy of a signal in a very low frequency band, average energy of a signal in a low frequency band, and average energy of a signal in a high frequency band.
10. The system of claim 5, wherein the non-linear features comprise at least one of a Poincare plot of a correlation between successive RR intervals in the heart signal and quantified long-range correlations generated by Detrended Fluctuation Analysis.
11. The system of claim 1, where the processor selects an optimal number of features comprising at least two features to provide a best classification accuracy for a combination of at least the first signature and the second signature.
12. The system of claim 11 further including a multilayer perceptron neural network for selecting the optimal number of features.
13. The system of claim 11, wherein the processor implements a k-nearest neighbor algorithm for selecting the optimal number of features.
15. The method of claim 14 further including generating an alarm indication responsive to generation of the second indication.
16. The method of claim 14, further comprising additional signatures comprising features of the heart signal, the features of the heart signal comprising at least one feature from the group consisting of chaotic features, catastrophe features, time domain features, frequency domain features, time-frequency domain features, and non-linear features.
17. The method of claim 16, wherein the topological features and the chaotic features comprise at least one of embedded dimensions of an attractor, fractional attractor dimensions, Correlation dimensions, Lyapunov exponent, Kolmogorov Entropy, Mutual Information, a proper delay for an attractor dimension, spectral density as a function of frequency, nonlinear chaotic measures, and a horizon for predictability.
18. The method of claim 16, wherein the time domain features comprise at least one of direct measurement of RR intervals, measurements of differences between RR intervals, a mean of all RR intervals, a standard deviation of all RR intervals, a square root of a mean of squares of differences between adjacent RR intervals and standard deviation of differences between adjacent RR intervals.
19. The method of claim 16, wherein the frequency domain features comprise low frequency band power spectral density features, high frequency band power spectral density features, very low frequency band power spectral density features, and ratio of low frequency band and high frequency band power spectral density.
20. The method of claim 16, wherein the time-frequency domain features comprise maximum amount of energy in a time window, minimum amount of energy in the time window, difference between maximum and minimum amount of energy between time windows, standard deviation between energy of time windows, total energy of a signal in a low frequency band, total energy of a signal in a high frequency band, average energy of a signal in a very low frequency band, average energy of a signal in a low frequency band, and average energy of a signal in a high frequency band.
21. The method of claim 16, wherein the non-linear features comprise at least one of a Poincare plot of a correlation between successive RR intervals in the heart signal and quantified long-range correlations generated by Detrended Fluctuation Analysis.
22. The method of claim 14 further comprising selecting an optimal number of features comprising at least two features to provide a best classification accuracy for a combination of at least the first signature and the second signature.
23. The method of claim 22, wherein the step of selecting further comprises selecting the optimal number of features using a k-nearest neighbor algorithm.
25. The system of claim 24, wherein the database is located within the wearable heart monitor.
26. The system of claim 24, where the processor selects an optimal number of features comprising at least two features to provide a best classification accuracy for a combination of at least the first signature and the second signature.
27. The system of claim 26 further including a multilayer perceptron neural network for selecting the optimal number of features.
28. The system of claim 26, wherein the processor implements a k-nearest neighbor algorithm for selecting the optimal number of features.
Cooperative Patent Classification codes for this invention. Click any code to explore related patents in that topic.
January 21, 2020
January 17, 2023
Browse 5M+ US patents with plain-English claim translations and AI-generated analysis.